Parent study
This study involved a secondary analysis of data from a recently completed observational study funded by the National Eye Institute. The parent study evaluated how communication between glaucoma patients and their ophthalmologists was associated with health outcomes. The parent study began enrollment in May 2009 and ended enrollment in November 2012; it included 15 ophthalmologists and 279 of their glaucoma patients. Patients and providers were recruited from 6 ophthalmology sites in the United States. Study data include videotapes of office visits and eye drop technique, patient interviews, medical record abstractions, and adherence data obtained via Medication Event Monitoring Systems (MEMS).
Eligibility Criteria
Patients were eligible for the parent study if they met the following criteria: a) age 18 or older; b) able to speak and understand English; c) were scheduled for a new glaucoma, glaucoma suspect, or glaucoma visit; d) mentally competent to participate; and e) not blind in both eyes (no perception of light).
Data Collection Procedures
Provider consent was obtained from the 15 ophthalmologists who participated in the study. All providers completed a demographic survey after providing consent. Regarding patient recruitment and enrollment, a research assistant (RA) explained the purpose of the study and obtained written consent from interested and eligible patients. Before the ophthalmologist entered the exam room, the RA set up a video camera, started recording and
43
exited the room. When the physician was finished with the visit, the RA re-entered the room and stopped the recorder. If the individual was diagnosed with glaucoma and prescribed a glaucoma medication, the RA conducted an interview with patients, videotaped their eye drop technique, and extracted the patient’s IOP from the medical record. The demographic information was collected during the interview with the patient. Patients were then given a large prescription vial with a MEMS cap in which to keep their eye drop prescription. One vial/MEMS cap was typically given for each separate glaucoma medication (up to 4 prescriptions). However, the number of MEMS caps given per patient was unique depending on the clinic. This was due to a temporary shortage of MEMS caps at some of the clinics during enrollment. Therefore, some patients may not have received caps for all of their glaucoma medications. The RA showed the patient how to use the MEMS cap and also gave the patient written instructions. The RA was responsible for calling the patient the next day to confirm that the eye drop containers were correctly placed in the prescription vials with the MEMS caps. The RA confirmed the follow-up appointment with the patient to collect adherence data. At the 4-6 week and 8-month follow-up appointments, the RA retrieved the MEMS device and scanned the data into a computer, which provided information on adherence for the 60 days following the baseline visit. The patients and physicians were unaware of the study hypotheses.
Transcript Generation
All videotapes were transcribed verbatim under the supervision of the principal investigator of the primary grant (not an eye care provider or associated with any of the centers involved). All identifiers were removed when the office videotapes were transcribed.
Transcript Coding
The transcripts were coded using a coding instrument developed specifically for this study. For the remainder of the dissertation, this coding instrument will be referred to as the ‘supplemental coding instrument’ because the parent study had its own coding instrument. An
44
initial draft of the supplemental coding instrument was developed by the author of the study and was refined and tested during the initial part of the study. The author of the study performed the majority of the coding with the supplemental instrument and was not blinded to the study hypotheses. A secondary coder was used to assess inter-coder reliability. Both coders were blinded to patient demographics. The second coder was blinded to the study hypotheses.
Measurement of Demographic Variables
This section describes the measurement of a) patient and physician demographics and b) medication and financial characteristics. The data came from two sources: 1) patient interview data collected at baseline and 2) medical record abstractions. Table 2 outlines the demographic variables and measures that were included in the analyses, including source, type and range.
Table 2: Patient, Medication, Financial, and Physician Variables, Source and Range
Variable Source Range
Patient Characteristics
Gender Patient Interview at Baseline 1=male; 0=female
Age (in years) Patient Interview at Baseline Continuous
Race/ethnicity Patient Interview at Baseline
1=White; 2=Asian; 3=African American; 4= Native American; 5=Hispanic; 6=Other
REALM health literacy Patient Interview at Baseline
1=eighth grade and below; 0= ninth grade and above
Severity of Disease (defect)
Visual field data
abstracted from Medical Record at Baseline
0=early; 1=moderate to severe
Comorbidities (patient self-reported)
Diabetes Patient Interview at
Baseline 1=yes; 0=no
High Blood Pressure Patient Interview at
Baseline 1=yes; 0=no
High Cholesterol Patient Interview at
45
Arthritis Patient Interview at Baseline 1=yes; 0=no
Hypothyroidism Patient Interview at
Baseline 1=yes; 0=no
Heart Disease Patient Interview at
Baseline 1=yes; 0=no
Depression Patient Interview at Baseline 1=yes; 0=no
Physician Characteristics
Gender Physician Questionnaire
at Baseline 1=male; 0=female
Race Physician Questionnaire
at Baseline 1=non-white; 0=white
Age (in years) Physician Questionnaire
at Baseline Continuous
Years Practicing Medicine Physician Questionnaire
at Baseline Continuous
Ophthalmologist Practice Type
Practice Type Physician Questionnaire
at Baseline
1=academic medical center; 0=private practice
Medication Characteristics Total Number of Glaucoma Medications
Medical Record
Abstraction at Baseline Discrete
New Glaucoma Medication User Medical Record
Abstraction at Baseline 1=yes; 0=no
Financial Pressures
Patient Annual Income Patient Interview at Baseline
1=less than $20,000; 2 = $20,000-$39,999; 3= $40,000-59,000; 4= $60,000-$79,999; 5= $80,000 or more; 6= don’t want to answer/don’t know
Does Patient Have Insurance Patient Interview at
Baseline 1=yes; 0=no
Type of Insurance Patient Interview at
Baseline
1=Medicaid;
2=Medicare; 3=Private; 4=Other
Does Patient Have Prescription Insurance
Patient Interview at
Baseline 1=yes; 0=no
Patient Indicates it is Hard to Pay for Glaucoma Medications
Patient Interview at
46 Patient Characteristics
The patient interview at baseline provided information about patient characteristics. Gender was recorded as a dichotomous variable, male or female. Age in years was measured as a continuous variable. The race of a patient was measured as a categorical variable: a) White, b) African American, c) Asian, d) Native American and e) Hispanic.
Another measure included in the study is the Rapid Estimate of Adult Literacy in Medicine (REALM). The REALM is a validated, rapid screening instrument that is designed to identify patients who have difficulty reading common medical and lay terms used in patient education materials.135
In the analysis, the REALM was dichotomized to eighth grade and below (REALM score of 0-60) or ninth grade and above reading level (REALM score of 61-66) since patients reading below ninth grade have trouble reading most patient education materials.135
Another group of variables included in the study was existence of chronic comorbid conditions. During the patient interview at baseline, patients reported if they had any of the following conditions: a) diabetes, b) high blood pressure, c) high cholesterol, d) arthritis, e) hypothyroidism, f) heart disease, or g) depression. These variables were dichotomized as yes/no.
Glaucoma severity was measured using the glaucoma staging system presented by Mills et al (2006).43 The four stages are: stage 1 (early glaucoma), stage 2 (moderate glaucoma), stage 3 (advanced glaucoma), and stage 4 (severe glaucoma). In the analysis, severity was evaluated both as a categorical variable and dichotomized variable. The dichotomized variable was separated as early versus moderate to severe glaucoma.
47 Physician Characteristics
Information concerning physician characteristics was obtained through the use of a physician questionnaire at baseline. Physician gender was recorded as female or male. The age (in years) of a physician at the beginning of the study was recorded as a continuous variable. Physician race was recorded as non-white or white. The length of time, in years, a physician has been practicing medicine was recorded as a continuous variable.
Ophthalmologist Practice Type
The ophthalmologist practice type was coded as a dichotomous variable, private or academic medical center.
Medication Characteristics
Information concerning medication characteristics was extracted from the patient medical record. The total number of glaucoma medications was recorded as a discrete variable. Combination medications were counted as one medication. Whether the patient was a new glaucoma medication user, was classified as a dichotomous variable, yes or no.
Financial Pressures
Information concerning financial pressures came from the baseline patient interview. The income level of the patient was recorded as a categorical variable with the following categories: a) less than $20,000, b) $20,000-39,999, c) $40,000-59,000, d) $60,000-79,999, e) $80,000 or more, f) don’t know/don’t want to answer. A patient’s health insurance was recorded two ways: 1) a dichotomous variable, yes or no and 2) a categorical variable, Medicaid,
48
Medicare, Private or Other. A patient’s availability of prescription insurance was recorded as a dichotomous variable, yes or no. The last financial pressure variable was patients indicate it is hard to pay for their glaucoma medications. This variable was dichotomized as yes or no.
Measurement of Communication Variables
Supplemental Coding Instrument
The supplemental coding instrument and coding rules are available in Appendix A. The source, range, and reliability of each of the variables are presented in Table 3 and described below. The supplemental coding instrument broke communication down into seven sections: 1) cost discussion and discussion initiator (patient or physician), 2) overall medication and other cost discussions, 3) provider behavior regarding cost discussions, 4) patient behavior regarding cost discussions, 5) insurance and drug cost, 6) samples, and 7) medication discussion.
Table 3: Communication Variables, Source, Range, and Reliability
Variable Source Range Reliability
ICCa Cost Discussion and Initiator
Cost Discussion Occurred Coding Tool 1=yes; 0=no 0.94
Initiator of Cost Discussion Coding Tool 1=Physician; 2=Patient
1.0
Overall Medication and Other Cost Discussion
Medication Cost Discussed Coding Tool 1=yes; 0=no 0.94
Patient Indicates Cost is a
Problem Coding Tool 1=yes; 0=no 0.88
Provider Behavior Physician Asks about a
Medication Cost Problem Coding Tool 1=yes; 0=no 1.0
Physician Proposes a Solution to
Potential Cost Problem Coding Tool 1=yes; 0=no 0.77
Physician Recommends Patient Work with Pharmacist to Lower Medication Cost
Coding Tool 1=yes; 0=no 1.0
Physician Asks How Much
Patient is Paying for Medications Coding Tool 1=yes; 0=no
100% agreement*
49 Physician Recommends Patient
Try 3-Month Supply to Reduce Cost
Coding Tool 1=yes; 0=no 1.0
Patient Behavior
Patient Changes Medication
Regime due to Cost Problem Coding Tool 1=yes; 0=no 0.79
Patient Discusses Coping Strategies Used in the Past to Deal with Cost Problems
Coding Tool 1=yes; 0=no 1.0
Insurance and Drug Cost Health and/or Prescription Drug
Insurance Discussed Coding Tool 1=yes; 0=no 0.93
Coinsurance and/or Copayment
is Discussed Coding Tool 1=yes; 0=no 0.88
Medication Assistance Programs
Discussed Coding Tool 1=yes; 0=no 1.0
Four Dollar Generics Discussed Coding Tool 1=yes; 0=no 1.0
Laser is Discussed as a Solution
to a Medication Cost Problem Coding Tool 1=yes; 0=no
100% agreement* Samples
Patient Requests Samples Coding Tool 1=yes; 0=no 100%
agreement*
Physician Provides Samples Coding Tool 1=yes; 0=no 0.91
Physician Discusses Being
Unable to Provide Samples Coding Tool 1=yes; 0=no
100% agreement* Medication Discussion
Term Brand or Generic is Used Coding Tool 1=yes; 0=no 1.0
Physician Initiates Discussion of
Brand or Generic Coding Tool 1=yes; 0=no 0.90
Physician Explains Difference
Between Brand and Generic Coding Tool 1=yes; 0=no 0.79
Patient Expresses Confusion Concerning Brand versus Generic
Coding Tool 1=yes; 0=no 1.0
Physician Says Generic isn’t
Available Coding Tool 1=yes; 0=no 0.79
Physician Offers to Write
Prescription for Generic Coding Tool 1=yes; 0=no 0.85
Physician Asks if Patient has a
Preference for Generic or Brand Coding Tool 1=yes; 0=no 1.0
Patient Expresses Preference for
Generic Coding Tool 1=yes; 0=no 1.0
Patient Expresses Preference for
Brand Coding Tool 1=yes; 0=no 1.0
50 Cost Discussion and Initiator
The first variable was number of cost discussions that occur zero, one, two, or three. A cost discussion was defined as a discussion of one topic relating to medication cost. Therefore, this discussion may continue throughout the transcript. For example, physicians may summarize discussions of medication cost at the end of the visit. In these cases, we would conclude that there was one cost discussion. The initiator of each cost discussion was recorded by selecting one of the following options: a) physician or b) patient. This variable was a categorical variable.
Overall Medication and Other Cost Discussions
The coding tool broadly identified what was discussed during the office visit by evaluating the following communication variables: a) glaucoma medication cost discussed during the baseline visit and b) patient indicates that cost is a problem during the baseline visit. The glaucoma medication cost variable was coded as yes if there was a discussion of medication cost, prescription drug insurance, and/or drug assistance programs. If there was no discussion of glaucoma medication cost in a transcript, then it was coded as no. The same process was used to determine whether the patient indicated that cost is a problem during the baseline visit. This variable was coded yes if a patient stated that he/she was having difficultly affording medications, medical visits, or other glaucoma related costs. The variable was also coded yes if the patient responded affirmatively when the physician asked him/her if cost is a problem.
Provider Behavior
The next section of the coding tool measured provider communication with the following variables: a) physician asks about a glaucoma medication cost problem, b) physician proposes
51
a solution to a potential glaucoma-related cost problem, c) physician recommends patient work with a pharmacist to lower glaucoma medication cost, d) physician asks how much patient is paying for glaucoma medications, and e) physician recommends patient try a 3-month supply of their glaucoma medications to reduce cost. All five variables were dichotomous and were coded yes or no. Physician asking about a glaucoma medication cost problem was coded yes if a physician asked patients if they were having problems affording their glaucoma medications. Physician proposing a solution to a potential glaucoma-related cost problem was coded yes if a physician provided a patient with a way to decrease glaucoma medication cost. Physician recommending the patient work with a pharmacist to lower glaucoma medication cost, was coded yes if the physician stated that the patient should speak to a pharmacist to lower glaucoma medication cost. Physician asking how much the patient is paying for glaucoma medications, was coded yes if the physician directly asked patients how much they paid out-of- pocket for their glaucoma medications. Physician recommending the patient try a 3-month supply to reduce glaucoma medication cost, was coded yes if the physician suggested the patient get a 3-month supply instead of a 1-month supply to save money on their glaucoma medications.
Patient Behavior
The next section of the coding tool measured patient communication with the following variables: a) patient had changed glaucoma medication regimen due to cost and b) patient discusses coping strategies used in the past to deal with cost problem. Both variables were dichotomous and were coded as yes or no. Patient changed medication regimen due to cost was coded yes if the patient discussed changing their glaucoma medication regimen because of cost problems at any point in the past. This included, decreasing the dosing schedule, skipping doses, decreasing the number of drops administered, and stopping medication use. Patient
52
discusses coping strategies used in the past to deal with a cost problem, was coded yes if the patient stated one or more coping methods they had used due to cost problems. This could include non-glaucoma related medications.
Insurance and Drug Cost
Insurance and drug cost communication were measured using the following variables: a) patient health insurance or prescription drug coverage discussed, b) coinsurance discussed or copayment discussed, c) medication assistance programs discussed, d) four dollar generics discussed, and e) laser discussed as a solution to a medication cost problem. All five variables were dichotomous and were coded as yes or no. Patient health insurance or prescription drug coverage discussed was coded as yes if the physician or patient talked about medical or drug insurance coverage. Coinsurance or copayment discussed was coded as yes if the patient or physician discussed coinsurance or copayment. Medication assistance programs discussed were coded as yes if the physician or patient talked about medication assistance programs , which in our definition included medication cards and other forms of assistance. Four dollar generics are discussed was coded as yes if the physician or patient talked about four dollar generics and/or what pharmacies sell these and/or what drugs are available for this offer. Laser discussed as a solution to a cost problem was coded yes if the physician discussed the possibility of laser surgery to reduce medication costs.
Samples
Communication regarding samples was measured with the following variables: a) patient requests samples, b) physician provides samples, and c) physician discusses being unable to provide samples. These three dichotomous variables were coded as yes or no. Patient requests samples was coded yes if the patient asked if samples are available for their glaucoma
53
medications or directly asked the physician for samples of glaucoma medications. Physician provides samples, was coded yes if the physician gave the patient glaucoma medication samples during the visit. Physician discusses being unable to provide samples was coded yes if the physician informed the patient that samples aren’t available or they are unable to provide them.
Medication Discussion
The coding tool measured communication concerning medications with the following variables: a) term brand or generic drug is used, b) physician initiates discussion of brand or generic, c) physician explains difference between brand and generic drugs, d) patient expresses confusion concerning brand versus generic drug, e) physician says generic isn’t available, f) physician offers to write prescription for generic, g) physician asks if patient has a preference for generic or brand, h) patient expresses preference for generic, and i) patient expresses preference for brand. All nine variables are dichotomous and were coded yes or no. The term brand or generic drug is used, was coded yes if the physician or patient used the words brand or generic when describing a glaucoma medication. Physician initiates discussion of brand or generic was coded yes if the physician was the first person to mention the term brand or generic. Physician explains difference between brand and generic drugs, was coded yes if the physician described the difference between brand and generic drugs. Patient expresses confusion concerning brand versus generic drug was coded yes if the patient indicated that they were confused about the difference between brand and generic drugs. Physician says generic is not available was coded yes if physicians told patients that a medication doesn’t have a